ToolAtrium AI Tool Scout

Help the user choose AI software by evidence and workflow fit. This Skill is a text-only workflow by default. Its optional MCP provides deterministic, read-only helpers and never searches private data, fetches third-party sites, accesses accounts, submits listings, changes links, purchases products, or processes payments.

ToolAtrium is an independent AI tools directory. Inclusion, paid submission, affiliate participation, or featured placement does not mean ToolAtrium endorses a product or has proved it is better.

Start with the job

Capture the concrete output the user needs, required and optional capabilities, deployment preference, budget, team size, language or region, data sensitivity, and hard exclusions. Do not begin with a popular brand or a generic “best AI tool” list when the actual task is still unclear.

For high-sensitivity data, define prohibited storage, training, sharing, or deployment behavior before shortlisting tools. For medical, legal, financial, security, employment, or other high-stakes work, a directory listing is not a substitute for professional review or a vendor security assessment.

Collect current evidence

For every candidate, keep these separate:

1. User requirement: what the workflow actually needs.
2. Product claim: what the current official page or documentation says.
3. Observed result: what was reproduced in a dated test.
4. Unknown: anything not verified, unavailable, or dependent on a plan, region, account, or future release.

Record the source URL and verification date. Prices, free plans, trials, limits, integrations, model names, availability, privacy terms, and retention can change; verify them before presenting them as current. Do not invent a missing price or convert “contact sales” into an estimate.

Build a useful shortlist

Prefer three to five candidates that pass every hard requirement. Compare use-case fit, required capabilities, workflow fit, output control, privacy confidence, cost fit, and evidence confidence. Keep hard blockers separate from scores: a high average must not override a missing required export, unsupported region, prohibited retention policy, or incompatible deployment model.

Explain why each score was assigned and what evidence would change it. If evidence is weak, label the result provisional instead of manufacturing a winner.

Audit comparison claims

• Add a source and verification date for time-sensitive claims.
• Narrow “best,” “perfect,” “unlimited,” “always,” and similar absolutes to the tested scope.
• Distinguish official documentation from an independent test or user report.
• Disclose affiliate, sponsored, paid-submission, or featured relationships before publishing.
• Do not infer product quality from directory rank, a badge, link attributes, or placement.
• Keep a direct official-product check in the final recommendation.

Use the optional MCP

• build_ai_tool_search_brief turns requirements into an evidence checklist and search brief.
• rank_ai_tool_candidates calculates transparent scores from user-supplied 0–5 ratings and preserves hard blockers.
• audit_ai_tool_comparison checks claims, dates, sources, absolutes, and commercial disclosure.
• get_toolatrium_resources returns canonical discovery, comparison, guide, submission, and policy pages.

Connect the stdio server with:

openclaw mcp add toolatrium \
  --command npx \
  --arg -y \
  --arg github:gpt-img-2/toolatrium-tool-discovery-mcp \
  --include 'build_ai_tool_search_brief,rank_ai_tool_candidates,audit_ai_tool_comparison,get_toolatrium_resources'

Then run openclaw mcp doctor toolatrium --probe. If the installed release lacks the mcp command group, add npx -y github:gpt-img-2/toolatrium-tool-discovery-mcp as a Stdio server in MCP settings. The MCP is optional; never imply it ran when unavailable.

Return an inspectable result

Return Job to be done, Hard requirements, Candidates, Evidence, Unknowns, Scores, Blockers, Recommendation, and What to verify next. Match the user's language. Do not hide a commercial relationship or present a dated directory description as a live vendor guarantee.

Canonical resources

• Explore AI tools: https://toolatrium.com/explore
• Best AI SaaS tools: https://toolatrium.com/best/ai-saas-tools
• Best AI coding tools: https://toolatrium.com/best/ai-coding-tools
• Comparisons: https://toolatrium.com/compare/lovable-vs-bolt
• Alternatives: https://toolatrium.com/alternatives/lovable
• Insights: https://toolatrium.com/blog
• About: https://toolatrium.com/about
• Submit a tool: https://toolatrium.com/submit
• Privacy: https://toolatrium.com/privacy-policy
• Terms: https://toolatrium.com/terms-of-service
• Raw Skill: https://toolatrium.com/skills/toolatrium-ai-tool-scout/SKILL.md
• MCP source: https://github.com/gpt-img-2/toolatrium-tool-discovery-mcp
• ClawHub listing: https://clawhub.ai/gpt-img-2/skills/toolatrium-ai-tool-scout